This paper describes a new word image generation model for word recognition. This model can generate a word image with likelihood based on linguistic knowledge, segmentation and character image. In the recognition process, first, the model generates the word image which approximates an input image best for each of a dictionary of possible words. Next, the model calculates the distance value between the input image and each generated word image. The efficiency of the proposed method as evaluated in an experiment using type-written museum archive card images. Results show that a recognition rate of 99.8% was obtained, compared with only 70.3%f or a recently published comparator algorithm.